WhoseDev AI Settings
WhoseDev AI Settings is the central place to configure how WhoseDev uses AI in your workspace. Open it from the workbench or Command Palette with WhoseDev: Open AI Hub (workbench.action.openWhoseDevAI).
The hub is organized into six sections:
| Section | What you configure |
|---|---|
| Overview | Current route, sign-in status, and quick links to common tasks |
| Models | Managed cloud models, composer visibility, and model availability for your plan |
| Providers | Bring-your-own-provider (BYOK) connections |
| Local AI | Ollama status, installed models, and offline model management |
| Routing | Auto routing, inline completion route, and response preference |
| Privacy & Usage | Context controls, route disclosure, and usage-related settings |
The status bar account menu (when signed in) shows plan name and Usage remaining for the selected WhoseDev workspace. See Plans and billing.
Models
Section titled “Models”The Models section lists models registered for your workspace, including WhoseDev-managed cloud models and locally installed Ollama models.
Managed models may appear more than once in the underlying registry (renderer and agent-host paths). The hub deduplicates entries so you see one row per logical model.
Models greyed out as unavailable are blocked by your plan, workspace policy, or backend availability — not a local configuration error.
For a deep dive on BYOK connections, see Bring your own provider. For local setup, see Local AI and Offline Models.
Providers
Section titled “Providers”Use Providers to add API keys for direct-provider routes. WhoseDev supports OpenAI, Anthropic, OpenRouter, Google Gemini, and Azure OpenAI. Keys are stored securely on this device; requests and billing go to the provider account you configure.
Local AI
Section titled “Local AI”The Local AI tab summarizes Ollama daemon status and links to WhoseDev: Manage Offline Models for guided setup, model pulls, and qualification results.
Routing
Section titled “Routing”The Routing section controls how WhoseDev selects models for managed requests.
Auto (managed cloud)
Section titled “Auto (managed cloud)”When Auto is selected for Composer Agent runs, WhoseDev profiles the privacy-masked request and current agent context, filters to capable models allowed by your plan, then optimizes for your response preference. The selected model stays pinned for the turn; static routing is the fallback when Auto is unavailable.
Response preference
Section titled “Response preference”Setting whosedev.com.routingPreference adjusts the latency/quality tradeoff for managed requests:
| Value | Behavior |
|---|---|
balanced |
Balance speed and quality (default) |
fast |
Prioritize faster responses |
quality |
Prioritize higher-quality responses |
Privacy & Usage
Section titled “Privacy & Usage”The Privacy & Usage section mirrors workspace privacy controls also documented in Privacy and usage, including what context is included with requests and whether web research is allowed.
Advanced model catalog
Section titled “Advanced model catalog”For the full managed model list beyond the default picker, run WhoseDev: Open Advanced Model Catalog (workbench.action.openAdvancedWhoseDevModelCatalog).